CN115102232A

CN115102232A - Method and system for judging low voltage ride through performance of wind power plant

Info

Publication number: CN115102232A
Application number: CN202210978130.7A
Authority: CN
Inventors: 张栋梁; 田鑫; 杨立超; 李永康; 刘宝柱; 李文升; 赵龙; 刘晓明; 杨斌; 杨思; 高效海; 王男; 张丽娜; 付一木; 魏佳; 魏鑫; 邱轩宇; 张玉跃; 袁振华; 程佩芬
Original assignee: North China Electric Power University; Economic and Technological Research Institute of State Grid Shandong Electric Power Co Ltd
Current assignee: North China Electric Power University; Economic and Technological Research Institute of State Grid Shandong Electric Power Co Ltd
Priority date: 2022-08-16
Filing date: 2022-08-16
Publication date: 2022-09-23
Anticipated expiration: 2042-08-16
Also published as: CN115102232B

Abstract

The invention relates to a method and a system for judging low voltage ride through performance of a wind power plant, belongs to the field of low voltage ride through, and is characterized in that a knowledge-driven time domain simulation method is adopted to simulate the change process of a power system before the fault clearing moment, after the fault clearing, the structure of the system does not change any more, the parameter change fluctuation is small, the wind power plant characteristic quantity at the fault clearing moment is utilized to predict the voltage related numerical value of a grid-connected point of the wind power plant through a data driving model, the data driving model is utilized to predict data and the voltage safety standard of the grid-connected point of the wind power plant, time domain simulation calculation is stopped in advance, and the judgment efficiency of whether the wind power plant can realize low voltage ride through is improved.

Description

Method and system for judging low voltage ride through performance of wind power plant

Technical Field

The invention relates to the field of low voltage ride through, in particular to a method and a system for judging low voltage ride through performance of a wind power plant.

Background

Power system stability is the ability of a power system to maintain stable operation after being disturbed by an accident. At present, the low voltage ride through capability of a wind power plant is a research hotspot, and the following requirements are met according to the requirements of GB/T19963-2011 technical provisions for accessing a wind power plant to a power system:

(1) when the voltage at the output end of the wind generating set falls to 20% of rated voltage, the wind generating set can guarantee continuous operation for 625ms without disconnection.

(2) The voltage of the output end of the wind generating set can be recovered to 90% of the rated voltage within 2s after the voltage falls off, and the wind generating set can guarantee continuous operation without off-grid.

When the problem of low voltage ride through of the wind power plant is researched, the time-domain simulation method is generally adopted to calculate the voltage change condition of the grid-connected point of the wind power plant along with time in detail. However, because the physical model of the wind power plant has high calculation dimension and strong variation fluctuation, the calculation speed is slow and the simulation time is long when the power system including the wind power plant is calculated by adopting a time domain simulation method. In order to solve the problem, some scholars propose to use neural network classification to judge whether the transient voltage of the wind field grid-connected point of the system can realize low voltage ride through. However, the method for judging whether low voltage ride through can be realized by adopting data driving still has the problems of less transient instability data, poor model interpretability, insufficient model judgment accuracy caused by severe parameter change when system faults occur and are cut off and the like.

Disclosure of Invention

The invention aims to provide a method and a system for judging the low voltage ride through performance of a wind power plant, so as to improve the judgment efficiency of whether the wind power plant can realize low voltage ride through.

In order to achieve the purpose, the invention provides the following scheme:

a method for judging low voltage ride through performance of a wind power plant comprises the following steps:

simulating the change process of the power system before the fault clearing time by adopting a knowledge-driven time domain simulation method, and calculating the lowest voltage of a grid-connected point of the wind power plant from the fault occurrence time of the power system to the fault clearing time and the characteristic quantity of the wind power plant at the fault clearing time;

inputting the characteristic quantity of the wind power plant at the fault removal moment into a data driving model, and predicting the final recovery voltage of a grid-connected point, the time required for the grid-connected point voltage to recover to 90% of a rated voltage and the lowest voltage after the fault removal;

judging whether the wind power plant meets the low voltage ride through test condition of the wind power converter or not according to the final recovery voltage of the grid-connected point, the time required for recovering the grid-connected point voltage to 90% of the rated voltage, the lowest voltage after fault removal and the lowest voltage of the wind power plant grid-connected point, and obtaining a first judgment result;

if the first judgment result shows that the wind power plant can realize low voltage ride through, judging that the wind power plant can realize low voltage ride through;

and if the first judgment result shows that the wind power plant cannot realize low voltage ride through, judging that the wind power plant cannot realize low voltage ride through.

Optionally, the simulating a power system change process before the fault removal time by using a knowledge-driven time domain simulation method, and calculating the lowest voltage of a grid-connected point of the wind farm from the power system fault occurrence time to the fault removal time and the wind farm characteristic quantity at the fault removal time specifically include:

acquiring an initial algebraic variable and an initial state variable of the power system at the simulation starting moment of a time domain simulation method, and setting an initial time step length and a maximum iteration number;

the initialization calculation times i are 0;

increasing the numerical value of the calculation frequency i by 1, and judging whether the calculation frequency i after the numerical value is increased is greater than the maximum iteration frequency or not to obtain a second judgment result;

if the second judgment result shows that the time is positive, reducing the initial time step length, and returning to the step that the preset calculation frequency i is 0;

if the second judgment result tableIf not, according to the initial algebraic variable and the initial state variable, using the system equation and the Jacobi matrix and adopting the forward Euler method to calculate the variable quantity delta y of the algebraic variable ⁽ⁱ⁾ And the amount of change Δ x of the state variable ⁽ⁱ⁾ ；

If | Δ y ⁽ⁱ⁾ | ≧ ε or | Δ x ⁽ⁱ⁾ If | > is more than or equal to epsilon, returning to the step of increasing the numerical value of the calculation times i by 1, and judging whether the calculation times i after the numerical value is increased are more than the maximum iteration times or not to obtain a second judgment result;

if | Δ y ⁽ⁱ⁾ |<ε and | Δ x ⁽ⁱ⁾ |<ε, then according to Δ y ⁽ⁱ⁾ And Δ x ⁽ⁱ⁾ Respectively updating an initial algebraic variable and an initial state variable, outputting the updated algebraic variable and state variable, and simultaneously recording the lowest voltage value of a grid-connected point of the wind power plant;

judging whether the simulation time after the next simulation time step is greater than the disturbance moment or not to obtain a third judgment result;

if the third judgment result shows that the first simulation time is the disturbance time, the first algebraic variable and the first state variable are respectively replaced by the first algebraic variable and the first state variable, and the first judgment result returns to the step of calculating the variable quantity delta y of the first algebraic variable by using a system equation and a Jacobian matrix according to the first algebraic variable and the first state variable and adopting a forward Euler method ⁽ⁱ⁾ And the amount of change Δ x of the state variable ⁽ⁱ⁾ "; the disturbance moment comprises a fault occurrence moment or a fault removal moment;

if the third judgment result shows no, replacing the initial algebraic variable and the initial state variable with the updated algebraic variable and the updated state variable respectively, returning to the step of calculating the variable quantity delta y of the algebraic variable by using a system equation and a Jacobian matrix and adopting a forward Euler method according to the initial algebraic variable and the initial state variable ⁽ⁱ⁾ And the amount of change Δ x of the state variable ⁽ⁱ⁾ ”；

Stopping simulation when the starting simulation time of the next simulation time step is equal to the fault removal time, and extracting the characteristic quantity of the wind power plant at the fault removal time;

and determining the minimum value from the lowest voltage value of the wind power plant grid-connection point at the power system fault occurrence moment to the lowest voltage of the wind power plant grid-connection point at the fault removal moment as the lowest voltage of the wind power plant grid-connection point from the power system fault occurrence moment to the fault removal moment.

Optionally, the obtaining of the initial algebraic variable and the initial state variable of the power system at the simulation start time of the time domain simulation method specifically includes:

and calculating the initial algebraic variable and the initial state variable of the power system at the simulation starting moment by adopting a Newton-Raphson method according to the power system parameters at the simulation starting moment.

Optionally, the algebraic variables include: the synchronous generator electromagnetic power, the grid-connected point voltage phase, the grid-connected point voltage amplitude, the wind speed of the wind motor and the electromagnetic power of the wind motor;

the state variables include: the method comprises the steps of (1) synchronizing the power angle of a generator, the rotating speed of a generator rotor and a wind motor, the current of a d shaft and a q shaft of the generator rotor and the blade angle of the generator;

the wind power plant characteristic quantity comprises: the method comprises the steps of wind power plant active power output, wind power plant reactive power output, wind power plant grid connection point voltage, fault duration and wind power plant rotor rotating speed.

Optionally, the data driving model is obtained by training a BP neural network through a training sample data set; the input of the training sample data set is wind power plant characteristic quantity under multiple scenes, and labels are final recovery voltage of a grid-connected point, time required for recovering the grid-connected point voltage to 90% of rated voltage and minimum voltage after fault removal.

Optionally, the determining, according to the final recovery voltage of the grid-connected point, the time required for recovering the grid-connected point voltage to 90% of the rated voltage, the minimum voltage after the fault is removed, and the minimum voltage of the grid-connected point of the wind farm, whether the wind farm meets the low voltage ride through test condition of the wind power converter is performed, so as to obtain a first determination result, specifically including:

if U is satisfied at the same time _end >＝0.9*U _N 、T _con +T _{Failure persistence} <＝2s、min[U _min 、U' _min ]>＝0.2*U _N If yes, the first judgment result shows yes;

if not satisfy U at the same time _end >＝0.9*U _N 、T _con +T _{Failure persistence} <＝2s、min[U _min 、U'min]>＝0.2*U _N If yes, the first judgment result shows no;

wherein, U _end For the final recovery voltage of the grid-connected point, T _con Time required for the grid-connected point voltage to recover to 90% of the rated voltage, T _{Failure persistence} For duration of fault, U _min Is the lowest voltage, U 'after fault removal' _min Is the lowest voltage of wind power plant grid-connected point, U _N Is a rated voltage.

A wind power plant low voltage ride through performance judgment system comprises:

the simulation module is used for simulating the change process of the power system before the fault removal moment by adopting a knowledge-driven time domain simulation method, and calculating the lowest voltage of a grid-connected point of the wind power plant from the fault occurrence moment of the power system to the fault removal moment and the characteristic quantity of the wind power plant at the fault removal moment;

the prediction module is used for inputting the characteristic quantity of the wind power plant at the fault removal moment into the data driving model, predicting the final recovery voltage of the grid-connected point, the time required for recovering the grid-connected point voltage to 90% of the rated voltage and the lowest voltage after the fault removal;

the judging module is used for judging whether the wind power plant meets the low voltage ride through test condition of the wind power converter or not according to the final recovery voltage of the grid-connected point, the time required for recovering the grid-connected point voltage to 90% of the rated voltage, the lowest voltage after fault removal and the lowest voltage of the grid-connected point of the wind power plant, and obtaining a first judgment result;

the first judgment module is used for judging that the wind power plant can realize low voltage ride through if the first judgment result shows that the wind power plant can realize low voltage ride through;

and the second judgment module is used for judging that the wind power plant cannot realize low voltage ride through if the first judgment result shows no.

Optionally, the simulation module specifically includes:

the system comprises a preset submodule and a time domain simulation submodule, wherein the preset submodule is used for acquiring an initial algebraic variable and an initial state variable of the power system at the simulation starting moment of the time domain simulation method, and setting an initial time step length and the maximum iteration times;

the initialization submodule is used for initializing the calculation times i to be 0;

the second judgment sub-module is used for increasing the numerical value of the calculation times i by 1, judging whether the increased calculation times i are greater than the maximum iteration times or not and obtaining a second judgment result;

a time step reducing submodule, configured to reduce an initial time step if the second determination result indicates yes, and return to the step "preset calculation time i is 0";

a variation calculation submodule, configured to calculate, if the second determination result indicates no, a variation Δ y of the algebraic variable according to the initial algebraic variable and the initial state variable by using a system equation and a jacobian matrix and using a forward euler method ⁽ⁱ⁾ And the amount of change Δ x of the state variable ⁽ⁱ⁾ ；

Calling submodule for if | Δ y ⁽ⁱ⁾ | ≧ ε or | Δ x ⁽ⁱ⁾ If | > is equal to or more than epsilon, calling a second judgment submodule;

update submodule for if | Δ y ⁽ⁱ⁾ |<ε and | Δ x ⁽ⁱ⁾ |<ε, then according to Δ y ⁽ⁱ⁾ And Δ x ⁽ⁱ⁾ Respectively updating an initial algebraic variable and an initial state variable, outputting the updated algebraic variable and state variable, and simultaneously recording the lowest voltage value of a grid-connected point of the wind power plant;

the third judgment submodule is used for judging whether the simulation time after the next simulation time step length is greater than the disturbance moment or not and obtaining a third judgment result;

a disturbance moment calculation submodule, configured to, if the third determination result indicates yes, adjust the initial time step length to update the simulation time after the next simulation time step length to the disturbance moment, replace the initial algebraic variable and the initial state variable with the updated algebraic variable and state variable, and call the variation calculation submodule; the disturbance moment comprises a fault occurrence moment or a fault removal moment;

the replacing submodule is used for respectively replacing the initial algebraic variable and the initial state variable with the updated algebraic variable and the updated state variable and calling the variable quantity calculating submodule if the third judgment result indicates that the algebraic variable and the initial state variable are not updated;

the simulation stopping submodule is used for stopping simulation when the starting simulation time of the next simulation time step length is equal to the fault removal time, and extracting the characteristic quantity of the wind power plant at the fault removal time;

and the minimum voltage simulation submodule is used for determining the minimum value from the minimum voltage value of the wind power plant grid-connected point at the power system fault occurrence moment to the minimum voltage of the wind power plant grid-connected point at the fault removal moment as the minimum voltage of the wind power plant grid-connected point from the power system fault occurrence moment to the fault removal moment.

the state variables include: the method comprises the steps of (1) synchronizing the power angle of a generator, the rotating speed of a rotor of the generator and a wind motor, the currents of a d shaft and a q shaft of the rotor of the generator and the blade angle of the generator;

the wind power plant characteristic quantity comprises: active output of the wind power plant, reactive output of the wind power plant, voltage of a grid connection point of the wind power plant, fault duration and rotor speed of the wind power plant.

Optionally, the determining module specifically includes:

a first judgment branch submodule for judging if U is satisfied simultaneously _end >＝0.9*U _N 、T _con +T _{Failure persistence} <＝2s、min[U _min 、U' _min ]>＝0.2*U _N If yes, the first judgment result shows yes;

a second branch-out submodule for determining if U is not satisfied simultaneously _end >＝0.9*U _N 、T _con +T _{Failure persistence} <＝2s、min[U _min 、U' _min ]>＝0.2*U _N If yes, the first judgment result shows no;

wherein, U _end For the final recovery voltage of the grid-connected point, T _con To be connected to the gridTime required for the point voltage to return to 90% of the rated voltage, T _{Failure persistence} For duration of fault, U _min Is the lowest voltage after fault removal, U' _min Is the lowest voltage of wind power plant grid-connected point, U _N Is a rated voltage.

According to the specific embodiment provided by the invention, the invention discloses the following technical effects:

the invention discloses a method and a system for judging low voltage ride through performance of a wind power plant.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.

FIG. 1 is a schematic diagram of a method for determining low voltage ride through performance of a wind farm according to an embodiment of the present invention;

fig. 2 is a flowchart of a method for determining low voltage ride through performance of a wind farm according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of training a data-driven model according to an embodiment of the present invention;

FIG. 4 is a block diagram of an IEEE14 node power system provided by an embodiment of the present invention;

FIG. 5 is a comparison graph of data-driven model prediction data and data obtained by a time domain simulation method according to an embodiment of the present invention; fig. 5 (a) is a final recovery voltage comparison diagram, fig. 5 (b) is a voltage recovery required time comparison diagram, fig. 5 (c) is a lowest voltage comparison diagram after the fault is removed, and fig. 5 (d) is a prediction result error diagram;

fig. 6 is a schematic diagram of an example of serially calculating a grid-connected point voltage of a wind farm by using a time domain simulation method and a data driving model according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

The embodiment of the invention provides a method for judging low voltage ride through performance of a wind power plant, aiming at the advantages and disadvantages of a time domain simulation method and a data driving method in the process of processing the transient power grid stability of a system. The method comprises the steps of firstly calculating a simulation process before the fault clearing time by using a time domain simulation method, after the fault clearing, enabling the system structure not to change any more and enabling parameter change fluctuation to be small, then inputting relevant characteristic quantity of the wind power plant grid-connected point voltage into a trained data driving model, and predicting the time required by the wind power plant grid-connected point final recovery voltage, the grid-connected point voltage to recover to 90% of rated voltage and the lowest voltage after the fault clearing through the data driving model. And comparing the predicted value with the requirements in GB/T19963-2011, judging whether the wind power plant can realize low voltage ride through, terminating time domain simulation calculation in advance, and improving the judgment efficiency of whether the wind power plant can realize low voltage ride through. The invention is a breakthrough and improvement on the existing method for judging whether the wind electric field can realize the low-voltage ride through, is an original design purpose of the invention and is different from the existing method for judging whether the wind electric field can realize the low-voltage ride through.

Referring to fig. 1 and fig. 2, a specific implementation process of the method for judging the low voltage ride through performance of the wind farm provided by the embodiment of the invention is as follows:

and S1, simulating the change process of the power system before the fault removal time by adopting a knowledge-driven time domain simulation method, and calculating the lowest voltage of the grid-connected point of the wind power plant from the fault occurrence time of the power system to the fault removal time and the characteristic quantity of the wind power plant at the fault removal time.

Illustratively, the specific process of this step is:

step 1, calculating the initial power flow of the power system at the moment by adopting a conventional method-Newton-Raphson method on the basis of system parameters at a certain moment. Since the newton-raphson method is not the focus of this patent, the calculation process briefly described here is: and forming a node admittance matrix, setting a voltage initial value, solving the unbalance amount of power and voltage, calculating a Jacobian matrix, solving a correction equation and correcting the voltage, and repeating the calculation steps until the system converges. The moment data is taken as time domain simulation method zero moment data, variable initial data is provided for subsequent time domain simulation gradual length calculation, and the read-in data specifically comprises data required by a dynamic differential equation set (generator transient state, sub-transient state potential change rule, rotor equation, excitation regulation system dynamic characteristic, speed regulation system dynamic characteristic, load dynamic characteristic, rectifier and inverter control behavior, other dynamic devices and the like) and a static algebraic equation set (power network equation, voltage equation of each line, voltage static characteristic equation of load, coordinate conversion equation and the like) in the calculation process of the time domain simulation method;

and 2, setting a time domain simulation calculation basic value, setting the initial calculation time step length of time domain simulation calculation as delta t, and adjusting the subsequent delta t to calculate more accurate system internal parameters at a specific moment. The amount of change Δ y of the algebraic variable (y) and the state variable (x) is calculated as shown in equation (4) using the system equations (equation (1), equation (2)) and the jacobian matrix (equation (3)) per time step ⁽ⁱ⁾ 、Δx ⁽ⁱ⁾ If | Δ y(i)|<ε and | Δ x ⁽ⁱ⁾ |<If epsilon is not satisfied, the system does not converge, and as shown in FIG. 1, the calculation times are recorded as i, and when the calculation times exceed the maximum iterative calculation times i in the set step length _max The calculation step size should be reduced to improve the calculation accuracy. According to the obtained variation quantity delta y ⁽ⁱ⁾ And Δ x ⁽ⁱ⁾ And updating the values of the variables over time. The system comprises a plurality of algebraic variables and state variables, wherein the algebraic variables specifically comprise the electromagnetic power of a synchronous generator, the voltage phase of a grid-connected point, the voltage amplitude of the grid-connected point, the wind speed of a wind motor, the electromagnetic power of the wind motor and the like; the state variables specifically comprise a power angle of the synchronous generator, the rotating speed of a rotor of the synchronous generator and a wind motor, d-axis and q-axis currents of a rotor of the wind generator, a blade angle of the wind generator and the like;

0＝f _n (x(t+Δt),y(t+Δt),f(t)) (1)

0＝g(x(t+Δt),y(t+Δt)) (2)

f and g in the formulas (1) and (2) respectively represent differential and algebraic equations, x and y are state variables and algebraic variables, and the equations are nonlinear; in the formula (3)

Is a Jacobian matrix of system states and algebraic variables, I _n For system feature matrices of the same size and order, F _x 、F _y 、G _x 、G _y Respectively, the rate of change of each parameter, F _x Partial derivatives of the state variables for the state variables, F _y For state variables versus algebraic variablesDerivation of the deviation, G _x Derivation of state variables for algebraic variables, G _y Calculating the partial derivative of the algebraic variable; the variation Δ x of the state variable and the algebraic variable is calculated stepwise by equation (4) ⁽ⁱ⁾ And Δ y ⁽ⁱ⁾ Updating the state change and the algebraic variable through an equation (5);

and 3, after the calculation of each time step is completed, scanning to check whether the power system is disturbed in the next time step. In the time domain simulation calculation process, the simulated time is t _k The simulation time after the next simulation time step is t _k+1 (t _k+1 ＝t _k + Δ t), time to failure t _{Occurrence of failure} If t is _{Occurrence of failure} <t _k+1 And then disturbance occurs in the next time domain simulation calculation step length, wherein the disturbance mainly comprises two moments of fault occurrence or fault removal. If disturbance occurs, adjusting the time step delta t to change the next calculation time into the disturbance occurrence time, and laying a cushion for accurately calculating each parameter of the system at the fault occurrence or fault removal time in the subsequent steps;

step 4, variable quantity delta x of state variable and algebraic variable obtained according to each time step in time domain simulation calculation process after fault occurrence ⁽ⁱ⁾ And Δ y ⁽ⁱ⁾ Updating the algebraic variable (y) and the state variable (x) after each time step, and outputting the numerical value of each variable obtained by calculation of a time domain simulation method after the time step is finished. Comparing the voltage of the grid-connected point of the wind band field obtained in each time step with corresponding data obtained in previous steps, and recording the lowest voltage U 'of the grid-connected point of the wind power plant in all time steps' _min Comparing the transient voltage with the predicted grid-connected point transient voltage of the wind power plant, and judging whether the wind power plant can realize low voltage ride through;

and 5, judging whether the knowledge-driven calculation method based on the time domain simulation method should be finished or not, and switching to a data driving method to predict the grid-connected point voltage of the wind power plant. And (4) checking whether the fault is removed, namely checking whether the structure of the system is changed in the subsequent time so as to cause drastic change of parameters in the system. If the fault is removed, the system structure does not change greatly, the voltage of the grid-connected point of the wind power plant does not drop and rise suddenly at the moment similar to the fault and removal, but changes slowly, the parameters in the system are relatively stable, and the data driving model can be used for predicting the voltage of the grid-connected point of the wind power plant; if the fault is not removed, namely the system still has structural change in the subsequent time and the parameters in the system have severe change, returning to the step 2 to continue calculating the change condition of the values in the next time step by using a time domain simulation method;

and 6, taking characteristic quantities related to the voltage of the grid-connected point of the wind power plant at the fault removal time as input of a data driving model, and selecting active output of the wind power plant, reactive output of the wind power plant, the voltage of the grid-connected point of the wind power plant, fault duration and the rotating speed of a rotor of the wind power plant as related characteristic quantities of the voltage of the grid-connected point of the wind power plant.

And step S2, inputting the wind power plant characteristic quantity at the fault removal moment into a data driving model, and predicting the final recovery voltage of the grid-connected point, the time required by the grid-connected point voltage to recover to 90% of the rated voltage and the lowest voltage after the fault removal.

Illustratively, on the basis of steps 1-6 illustrated in step S1, the specific implementation process of step S2 is:

and 7, constructing a data driving model, and acquiring simulation data related to the grid-connected point voltage of the wind power plant in multiple scenes by setting random fault clearing time, random load and random generator output. Corresponding to the characteristic quantity related to the voltage of the grid-connected point of the wind power plant selected in the step 6, extracting the characteristic quantities related to the voltage of the grid-connected point of the wind power plant (active power output, reactive power output, grid-connected point voltage, fault duration and rotor speed of the wind power plant) in a large amount of simulation data as the input of a data driving model, and extracting the recovery voltage U from the grid-connected point in sample data _end And the time T required by the voltage of the grid-connected point to recover to 90 percent of rated voltage _con Lowest voltage U after fault removal _min The data is used as the output of the data-driven model, and the data-driven prediction model is trained by the BP neural network, as shown in FIG. 3.

Step 8, the data extracted in the step 6 are used as input, and the final recovery voltage U of the grid-connected point is predicted through the trained data driving model in the step 7 _end Grid-connected point power supplyTime T required for pressure to return to 90% of rated voltage _con And the lowest voltage U after fault removal _min And therefore, the general change trend of the grid-connected point voltage of the wind power plant after the fault is cut off is described.

And step S3, judging whether the wind power plant meets the low voltage ride through test condition of the wind power converter or not according to the final recovery voltage of the grid-connected point, the time required for recovering the grid-connected point voltage to 90% of the rated voltage, the lowest voltage after fault removal and the lowest voltage of the grid-connected point of the wind power plant, and obtaining a first judgment result.

And step S4, if the first judgment result shows that the wind power plant can realize low voltage ride through.

And step S5, if the first judgment result shows no, judging that the wind power plant can not realize low voltage ride through.

In one example, the predicted approximate trend of the grid-connected point voltage of the wind power plant is compared with a voltage limit value required by a low voltage ride through test of a wind power converter in GB/T19963-2011 technical Specification for connecting a wind power plant to a power system, if U is _end >＝0.9*U _N 、T _con +T _{Failure persistence} <＝2s、min[U _min 、U' _min ]>＝0.2*U _N And meanwhile, the requirement is met, the wind power plant can realize low voltage ride through, otherwise, the wind power plant cannot realize low voltage ride through.

The validity of the invention is verified by taking the structure of the IEEE14 node power system shown in FIG. 4 as an example.

Knowledge and data hybrid driving-based wind power plant grid-connected point transient voltage prediction

According to the wind power plant grid-connected point transient voltage prediction method based on knowledge and data hybrid driving, a data and knowledge hybrid driving mode is a serial mode, the change condition of a system before fault removal is calculated by a knowledge-driven time domain simulation method, and then a data driving model is adopted to predict the wind power plant grid-connected point voltage. The effectiveness of the invention is verified by using an IEEE14 node power system (as shown in figure 4) with wind power, wherein a wind power grid connection point is at a bus1, the fault type is a three-phase short-circuit fault, the fault occurs between a bus 6 and a bus 12 and is close to the bus 6, the fault occurs at 0.1s, and a breaker is disconnected to remove the fault at 0.25 s. In fig. 4, Bus01 to Bus14 represent Bus1 to Bus 14.

1. Because the internal structure of the system changes at the fault occurrence time and the fault removal time, the parameters in the system change violently, and if the data driving model is adopted for calculation, a large error exists, so that before the fault removal time, in order to deal with the violent change fluctuation of the system parameters, a more accurate calculation value is obtained, and the parameter change condition at the time before the fault removal is calculated by adopting a knowledge-driven time domain simulation method. In the calculation example, the time before the fault removal time (0.25s) is calculated by a time domain simulation method.

2. And after the fault is removed, the system structure is not changed any more, the system parameter change is stable, at the moment, the related characteristic quantity of the last step of the time domain simulation method is used as the input of the data driving model to predict the voltage of the grid-connected point of the wind power plant, and therefore the problem that the data driving method cannot well learn the scene of severe data fluctuation in the calculation process is solved. After the 0.25s fault is removed in the calculation example, the data at the moment calculated by the time domain simulation method is extracted, and the subsequent voltage change condition is predicted by using a data driving model.

3. And finally, recording the lowest voltage U 'of a wind power plant grid-connected point in the time domain simulation method' _min And the final recovery voltage U of the grid-connected point is obtained by predicting the data driving model _end Time T required for recovering grid-connected point voltage to 90% of rated voltage _con Lowest voltage U after fault removal _min As shown in fig. 5. Comparing with voltage limit value required by low voltage ride through test of wind power converter in GB/T19963-2011 technical Specification for accessing wind power plant into electric power system, if U is _end >＝0.9*U _N 、T _con +T _{Failure persistence} <＝2s、min[U _min 、U' _min ]>＝0.2*U _N And meanwhile, the requirement is met, the wind power plant can realize low voltage ride through, otherwise, the wind power plant cannot realize low voltage ride through.

The method provided by the invention fully combines the advantages of the knowledge driving and data driving methods by using a serial structure, solves the error problem caused by severe parameter fluctuation by using the knowledge driving, solves the problem of overlong time of the time domain simulation method by using the data driving, terminates the time domain simulation calculation in advance and improves the calculation efficiency.

Data extraction based on time domain simulation method

The data of the data driving model in the invention is derived from a time domain simulation method, and the characteristic quantity with strong correlation with the grid-connected point voltage of the wind power plant in the time domain simulation method fault clearing moment data is taken as input, such as: active power output, reactive power output, grid-connected point voltage, fault duration and rotor rotation speed of the wind power plant.

1. Time domain simulation computation

The time domain simulation calculation firstly needs to calculate the initial load flow, the initial load flow result is set as the initial value in the time domain simulation method, and the time step-by-time step iterative calculation is carried out subsequently to obtain the time-varying condition of each variable.

(1) Firstly, calculating the power flow at a certain moment, calculating the initial power flow of the power system at the moment on the basis of the system parameters at the moment, and setting the variable values at the moment as the initial values of the variables in the time domain simulation calculation.

(2) And taking the calculated basic load flow moment data as time-domain simulation method zero moment data, providing variable initial data for subsequent time-domain simulation step-by-step long calculation, and performing time-domain simulation calculation according to the time step delta t.

2. Extracting characteristic quantity related to wind power plant grid-connected point voltage at fault removal moment

Calculating two time points with severe structural changes of a system, namely fault occurrence and fault removal, by a time domain simulation method, and extracting characteristic quantities related to the grid-connected point voltage of the wind power plant at the fault removal time, wherein the characteristic quantities related to the grid-connected point voltage of the wind power plant are selected in the invention, and as shown in FIG. 3: the method comprises the steps of wind power plant active power output, wind power plant reactive power output, wind power plant grid connection point voltage, fault duration and wind power plant rotor rotating speed. The per unit value of active output of the wind power plant, the per unit value of reactive output of the wind power plant, the per unit value of voltage of a grid-connected point of the wind power plant, the fault duration and the rotating speed radian of a rotor of the wind power plant at the moment of fault removal extracted in the calculation example are respectively 3.18, 0.42, 0.89, 0.15s and 0.8 rad/s.

Additionally extracting time domainIn the simulation calculation process, the lowest voltage U 'of grid-connected point of wind power plant from fault occurrence to fault removal moment' _min And subsequently comparing the transient voltage with the predicted wind power plant grid-connected point transient voltage, and judging whether the wind power plant can realize low voltage ride through. U 'in example' _min The per unit value is 0.65.

Third, construction and application of data driving model

1. Construction of data-driven models

A large amount of real data are obtained through simulation, as shown in FIG. 3, characteristic quantities, namely active output of a wind power plant, reactive output of the wind power plant, voltage of a grid-connected point of the wind power plant, fault duration and rotor speed of the wind power plant, of the fault removal moment and related to voltage of the grid-connected point of the wind power plant are extracted from a time domain simulation method and are used as input of a data driving model, and finally recovery voltage U of the grid-connected point is used _end And the time T required by the voltage of the grid-connected point to recover to 90 percent of rated voltage _con And the lowest voltage U after fault removal _min A data-driven model is constructed and trained for output. In the calculation example, in order to construct a data driving model, 700 sets of simulation data samples are obtained by setting random fault time, random wind speed and random load, related data of the samples are extracted, a BP neural network (the hidden layer is 1 layer, wherein the number of neurons is 6) is adopted to train the data driving model, and various prediction data errors of the obtained data driving model are shown in FIG. 5.

2. Application of data-driven model

The trained data driving model is connected with a time domain simulation method in a serial mode, as shown in fig. 6, the calculation process of the time domain simulation method is firstly carried out, the voltage related data of the grid-connected point of the wind power plant at the fault clearing moment of the time domain simulation method is used as the input data of the data driving model, and the voltage related value (including the final recovery voltage U of the grid-connected point) of the wind power plant is predicted _end Time T required for recovering grid-connected point voltage to 90% of rated voltage _con And the lowest voltage U after fault removal _min ). Calculating data obtained by a time-domain simulation method at the moment of fault removal (the wind power plant active output per unit value at the moment of fault removal is 3.18, the wind power plant reactive output per unit value is 0.42, the wind power plant grid-connected point voltage per unit value is 0.89, the fault duration is 0.15s and the wind power plantRotor speed 0.8rad/s) input data driving model, and predicting final recovery voltage U of grid-connected point _end Time T required for recovering grid-connected point voltage to 90% of rated voltage _con And the lowest voltage U after fault removal _min The per unit values are 1.063, 0.0384s, 0.8867, respectively.

Fourth, judging low voltage ride through capability

As shown in fig. 6, the predicted data of the data driving mode is compared with the voltage limit value required by the low voltage ride through test of the wind power converter in GB/T19963-2011 technical provisions for accessing the power system to the wind farm, and the predicted voltage variation trend is depicted in the three directions of the recovery voltage threshold, the recovery time threshold, and the minimum voltage threshold in the present invention.

1. Recovery voltage threshold

Recovery voltage threshold requirement, i.e. U _end >＝0.9*U _N And requiring the final recovery voltage of the wind power plant to be above 90% of the rated voltage.

2. Recovery time threshold

Recovery time threshold requirement, i.e. T _con +T _{Failure persistence} <The predicted recovery time after the wind farm fault occurs plus the fault duration is required to be less than the maximum time that the wind farm can support 2 s.

3. Minimum voltage threshold

Minimum voltage threshold requirement, i.e. min U _min 、U' _min ]>＝0.2*U _N Calculating and recording the lowest voltage U 'of a grid-connected point of the wind power plant from the time point when the fault occurs to the time point when the fault is cut off by using a time domain simulation method' _min And the value and the data driving model predict the lowest voltage U of the wind power plant grid-connected point after the fault is removed _min Should be greater than the minimum voltage threshold by 20% of the rated voltage.

If the three conditions are met at the same time, the wind power plant is considered to be capable of realizing low voltage ride through, that is, as shown in fig. 6, the real voltage is within the safety range of the low voltage ride through test requirement voltage specification of the wind power converter in the technical specification of wind power plant access power system of GB/T19963-2011.

As shown in FIG. 6, the final recovery voltage U in the example _end Per unit value of 1.063Greater than 0.9; time T required for restoring grid-connected point voltage to 90% of rated voltage _con 0.0384s, and a fault duration T _{Failure persistence} 0.1884s in total is less than 2 s; lowest voltage U after fault removal _min U 'in time domain simulation method with per unit value of 0.8867' _min The unit value is 0.65, min 0.65, 0.8867]>0.2; the three judgment conditions are all satisfied, so the wind electric field can realize low voltage ride through in the example.

The invention has the following advantages:

1. according to the method, a serial mode of hybrid driving of a data model and a physical model is utilized, a time domain simulation method is firstly used for calculating a simulation process before a fault clearing moment, after the fault clearing, the structure of the system does not change any more, the parameter change fluctuation is small, and at the moment, the data driving model is used for predicting the voltage related value of a grid-connected point of the wind power plant. The time domain simulation calculation can be terminated in advance, and the efficiency of judging whether the wind power plant can realize low voltage ride through is improved.

2. According to the method, a data driving model is constructed to predict the final recovery voltage of the wind power plant grid-connected point, the time required for the wind power plant grid-connected point voltage to recover to 90% of rated voltage and the minimum voltage of the wind power plant grid-connected point after fault removal by extracting the related characteristic quantities of the wind power plant grid-connected point voltage (active power output of the wind power plant, reactive power output of the wind power plant, wind power plant grid-connected point voltage, fault duration and wind power plant rotor rotating speed).

3. The method utilizes the comparison of the data driving model prediction data and the wind power grid-connected point voltage safety standard to judge whether the wind electric field can realize low voltage ride through.

The embodiment of the invention also provides a system for judging the low voltage ride through performance of the wind power plant, which comprises the following steps:

the judging module is used for judging whether the wind power plant meets the low voltage ride through test condition of the wind power converter or not according to the final recovery voltage of the grid-connected point, the time required for recovering the voltage of the grid-connected point to 90% of the rated voltage, the lowest voltage after fault removal and the lowest voltage of the grid-connected point of the wind power plant, and obtaining a first judgment result;

The simulation module specifically comprises:

the second judgment submodule is used for increasing the numerical value of the calculation frequency i by 1, judging whether the calculation frequency i after the numerical value is increased is greater than the maximum iteration frequency or not and obtaining a second judgment result;

a variation calculating submodule, configured to calculate a variation Δ y of the algebraic variable according to the initial algebraic variable and the initial state variable, using a system equation and a jacobian matrix, and using a forward eulerian method if the second determination result indicates no ⁽ⁱ⁾ And the amount of change Δ x of the state variable ⁽ⁱ⁾ ；

update submodule for if | Δ y ⁽ⁱ⁾ |<ε and | Δ x ⁽ⁱ⁾ |<Epsilon, thenAccording to Δ y ⁽ⁱ⁾ And Δ x ⁽ⁱ⁾ Respectively updating an initial algebraic variable and an initial state variable, outputting the updated algebraic variable and state variable, and simultaneously recording the lowest voltage value of a grid-connected point of the wind power plant;

a disturbance moment calculation submodule, configured to adjust an initial time step if the third determination result indicates that the simulation time after the next simulation time step is the disturbance moment, replace the initial algebraic variable and the initial state variable with the updated algebraic variable and state variable, and call the variation calculation submodule; the disturbance moment comprises a fault occurrence moment or a fault removal moment;

the replacing submodule is used for respectively replacing the initial algebraic variable and the initial state variable with the updated algebraic variable and state variable and calling the variable quantity calculating submodule if the third judgment result shows that the initial algebraic variable and the initial state variable are not replaced;

and the minimum voltage simulation submodule is used for determining the minimum value from the lowest voltage value of the wind power plant grid-connected point at the power system fault occurrence moment to the lowest voltage of the wind power plant grid-connected point at the fault removal moment as the lowest voltage of the wind power plant grid-connected point at the power system fault occurrence moment to the fault removal moment.

The algebraic variables include: the synchronous generator electromagnetic power, the grid-connected point voltage phase, the grid-connected point voltage amplitude, the wind speed of the wind motor and the electromagnetic power of the wind motor;

The judging module specifically comprises:

a second judging branch submodule for satisfying U if not simultaneously _end >＝0.9*U _N 、T _con +T _{Failure persistence} <＝2s、min[U _min 、U' _min ]>＝0.2*U _N If yes, the first judgment result shows no;

wherein, U _end For final recovery of voltage at grid connection point, T _con Time required for the grid-connected point voltage to recover to 90% of the rated voltage, T _{Failure persistence} For duration of fault, U _min Is the lowest voltage, U 'after fault removal' _min Minimum voltage of wind power plant grid-connected point, U _N Is a rated voltage.

The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.

The principle and the embodiment of the present invention are explained by applying specific examples, and the above description of the embodiments is only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims

1. A method for judging low voltage ride through performance of a wind power plant is characterized by comprising the following steps:

inputting the characteristic quantity of the wind power plant at the fault removal moment into a data driving model, and predicting the final recovery voltage of a grid-connected point, the time required for recovering the grid-connected point voltage to 90% of rated voltage and the lowest voltage after the fault removal;

2. The method according to claim 1, wherein the simulating the power system change process before the fault removal time by using a knowledge-driven time domain simulation method, and calculating the wind farm grid-connected point minimum voltage from the power system fault occurrence time to the fault removal time and the wind farm characteristic quantity at the fault removal time specifically comprise:

the initialization calculation times i are 0;

if the second judgment result shows no, according to the initial algebraic variable and the initialCalculating the variable quantity delta y of the algebraic variable by using a system equation and a Jacobian matrix and adopting a forward Euler method ⁽ⁱ⁾ And the amount of change Δ x of the state variable ⁽ⁱ⁾ ；

If | Δ y ⁽ⁱ⁾ | ≧ ε or | Δ x ⁽ⁱ⁾ If | ≧ epsilon, returning to the step of increasing the numerical value of the calculation times i by 1, and judging whether the calculation times i after the numerical value increase is greater than the maximum iteration times to obtain a second judgment result;

if | Δ y ⁽ⁱ⁾ < ε and | Δ x ⁽ⁱ⁾ If | < ε, then according to Δ y ⁽ⁱ⁾ And Δ x ⁽ⁱ⁾ Respectively updating an initial algebraic variable and an initial state variable, outputting the updated algebraic variable and state variable, and simultaneously recording the lowest voltage value of a grid-connected point of the wind power plant;

if the third judgment result shows no, replacing the initial algebraic variable and the initial state variable with the updated algebraic variable and state variable respectively, and returning to the step of calculating the variation delta y of the algebraic variable by using a system equation and a Jacobian matrix according to the initial algebraic variable and the initial state variable and adopting a forward Eulerian method ⁽ⁱ⁾ And the amount of change Δ x of the state variable ⁽ⁱ⁾ ”；

and determining the minimum value from the lowest voltage value of the wind power plant grid-connected point at the power system fault occurrence moment to the lowest voltage value of the wind power plant grid-connected point at the fault removal moment as the lowest voltage of the wind power plant grid-connected point at the power system fault occurrence moment to the fault removal moment.

3. The method according to claim 2, wherein the obtaining of the initial algebraic variables and the initial state variables of the power system at the simulation start time of the time-domain simulation method specifically comprises:

4. The method of claim 2, wherein the algebraic variables comprise: the synchronous generator electromagnetic power, the grid-connected point voltage phase, the grid-connected point voltage amplitude, the wind speed of the wind motor and the electromagnetic power of the wind motor;

5. The method of claim 1, wherein the data-driven model is obtained by training a BP neural network with a set of training sample data; the input of the training sample data set is wind power plant characteristic quantity under multiple scenes, and labels are final recovery voltage of a grid-connected point, time required for recovering the grid-connected point voltage to 90% of rated voltage and minimum voltage after fault removal.

6. The method according to claim 1, wherein the step of judging whether the wind farm meets a low voltage ride through test condition of the wind power converter or not according to the final recovery voltage of the grid-connected point, the time required for recovering the grid-connected point voltage to 90% of the rated voltage, the lowest voltage after fault removal and the lowest voltage of the wind farm grid-connected point is performed to obtain a first judgment result, specifically comprises the steps of:

if U is satisfied at the same time _end ＞＝0.9*U _N 、T _con +T _{Failure persistence} ＜＝2s、min[U _min 、U' _min ]＞＝0.2*U _N If yes, the first judgment result shows yes;

if not simultaneously satisfy U _end ＞＝0.9*U _N 、T _con +T _{Failure persistence} ＜＝2s、min[U _min 、U' _min ]＞＝0.2*U _N If yes, the first judgment result shows no;

wherein, U _end For final recovery of voltage at grid connection point, T _con Time required for the grid-connected point voltage to recover to 90% of the rated voltage, T _{Failure persistence} For duration of fault, U _min Is the lowest voltage, U 'after fault removal' _min Is the lowest voltage of wind power plant grid-connected point, U _N Is a rated voltage.

7. A wind power plant low voltage ride through performance judgment system is characterized by comprising:

8. The system of claim 7, wherein the simulation module specifically comprises:

update submodule for if | Δ y ⁽ⁱ⁾ < ε and | Δ x ⁽ⁱ⁾ If | < ε, then according to Δ y ⁽ⁱ⁾ And Δ x ⁽ⁱ⁾ Respectively updating an initial algebraic variable and an initial state variable, outputting the updated algebraic variable and state variable, and simultaneously recording the lowest voltage value of a grid-connected point of the wind power plant;

9. The system of claim 8, wherein the algebraic variables comprise: the synchronous generator electromagnetic power, the grid-connected point voltage phase, the grid-connected point voltage amplitude, the wind speed of the wind motor and the electromagnetic power of the wind motor;

10. The system according to claim 7, wherein the determining module specifically includes:

first judgment branchModule for if U is satisfied simultaneously _end ＞＝0.9*U _N 、T _con +T _{Failure persistence} ＜＝2s、min[U _min 、U' _min ]＞＝0.2*U _N If yes, the first judgment result shows yes;

a second judging branch submodule for satisfying U if not simultaneously _end ＞＝0.9*U _N 、T _con +T _{Failure persistence} ＜＝2s、min[U _min 、U' _min ]＞＝0.2*U _N If yes, the first judgment result shows no;